Title:	CIRCUMPLEX SORT: A TWO-PHASE METHOD FOR REVEALING CIRCUMPLEX DATA PATTERNS IN REORDERABLE MATRICES
Author(s):	Miguel Mechi Naves Rocha, Pedro Kretikouski Roque Jr. and Celmar Guimarães da Silva
ISBN:	978-989-8533-66-1
Editors:	Yingcai Xiao and Ajith P. Abraham
Year:	2017
Edition:	Single
Keywords:	Reorderable matrices, Circumplex data pattern, seriation, reordering algorithms, heatmap
Type:	Full Paper
First Page:	181
Last Page:	188
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	Reorderable matrices are the basis of heatmaps and other tabular displays in which row and column permutations are possible. Matrix reordering algorithms try to provide a permuted version of these matrices, such that hidden patterns become visible for users. However, most of them fail on revealing Circumplex pattern, due to high execution time or to low quality of output matrix. In order to overcome this difficulty, we present a matrix reordering method (Circumplex Sort), whose goal is to permute a matrix in order to reveal a Circumplex pattern. In a first phase, our method reorders input matrix rows according to the index-weighted means of the values of each column. After that, the method creates and concatenates two stacks of rows based on a same initial row; for each row obtained according to the row order of the first phase, the method pushes this row at the stack whose top is most similar to it. Circumplex Sort runs the same two phases for columns. Our experiments with synthetic matrices revealed that our approach produces high-quality results (in terms of evaluation functions and runtime) when this pattern is hidden in data matrix. Besides, we present a real world dataset that exemplifies the use of Circumplex Sort for revealing Circumplex pattern.
	Go Back